Urban disaster management is currently limited, because it treats typhoon disasters in isolation and does not consider the cascading effects when multiple disasters interact. In addition, the current literature does not clearly recognize cross-sector collaboration networks, which is a countermeasure for reducing the risks of urban disasters and maintaining sustainable development. This study adopted a complex network approach to examine the cascading effects of typhoon disasters and to identify the types of hazard events and vulnerability paths for urban areas. Cross-sector collaboration networks were constructed to identify the network parameters that facilitate collaborative behaviors among different sectors. Using a comparative analysis of cases (Japan and the Philippines), two weighted complex networks were constructed to simulate disaster cascading effects in urban systems. The results can be summarized as four aspects: (1) For the node degree centrality, rainstorms had the highest values in the typhoon disaster network of Japan and the Philippines. (2) Considering the node betweenness centrality, rainstorms played an extremely crucial role in typhoon disasters with cascading effects in Japan. In contrast, landslides were the most difficult disasters to mitigate in the Philippines. (3) After the analysis of node and edge removal strategy, the global efficiency of network decreased from 0.092 and 0.084 to 0.064, indicating all intentional removal strategies led to better disaster management performance than random removal strategies. (4) The results of the Exponential Random Graph Model (ERGM) indicate that network structures and node attributes were the main factors affecting the formation of cross-sector collaborative networks. These results provide information and increase understanding about the typhoon disasters with cascading effects in urban systems. This study furnishes effective disaster prevention and reduction strategies for urban managers to improve urban sustainability and resilience.